Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks

Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks

This paper explores the value of <i>weak-ties</i> in classifying academic literature with the use of graph convolutional neural networks. Our experiments look at the results of treating <i>weak-ties</i> as if they were <i>strong-ties</i> to determine if that assum...

Full description

Bibliographic Details
Main Authors: Neda H. Bidoki, Alexander V. Mantzaris, Gita Sukthankar
Format: Article
Language:English
Published: MDPI AG 2020-05-01
Series:Machine Learning and Knowledge Extraction
Subjects:
graph convolutional neural networks
weak ties
social networks
collective classification
Online Access:https://www.mdpi.com/2504-4990/2/2/8
id doaj-99ee9bf3a7014577b5e879d8b2c0488f
record_format Article
spelling doaj-99ee9bf3a7014577b5e879d8b2c0488f2020-11-25T03:10:57ZengMDPI AGMachine Learning and Knowledge Extraction2504-49902020-05-012812514610.3390/make2020008Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural NetworksNeda H. Bidoki0Alexander V. Mantzaris1Gita Sukthankar2Department of Computer Science, University of Central Florida (UCF), Orlando, FL 32816, USADepartment of Statistics and Data Science, University of Central Florida (UCF), Orlando, FL 32816, USADepartment of Computer Science, University of Central Florida (UCF), Orlando, FL 32816, USAThis paper explores the value of <i>weak-ties</i> in classifying academic literature with the use of graph convolutional neural networks. Our experiments look at the results of treating <i>weak-ties</i> as if they were <i>strong-ties</i> to determine if that assumption improves performance. This is done by applying the methodological framework of the Simplified Graph Convolutional Neural Network (SGC) to two academic publication datasets: Cora and Citeseer. The performance of SGC is compared to the original Graph Convolutional Network (GCN) framework. We also examine how node removal affects prediction accuracy by selecting nodes according to different centrality measures. These experiments provide insight for which nodes are most important for the performance of SGC. When removal is based on a more localized selection of nodes, augmenting the network with both <i>strong-ties</i> and <i>weak-ties</i> provides a benefit, indicating that SGC successfully leverages local information of network nodes.https://www.mdpi.com/2504-4990/2/2/8graph convolutional neural networksweak tiessocial networkscollective classification
collection DOAJ
language English
format Article
sources DOAJ
author Neda H. Bidoki
Alexander V. Mantzaris
Gita Sukthankar
spellingShingle Neda H. Bidoki
Alexander V. Mantzaris
Gita Sukthankar
Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
Machine Learning and Knowledge Extraction
graph convolutional neural networks
weak ties
social networks
collective classification
author_facet Neda H. Bidoki
Alexander V. Mantzaris
Gita Sukthankar
author_sort Neda H. Bidoki
title Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
title_short Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
title_full Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
title_fullStr Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
title_full_unstemmed Exploiting Weak Ties in Incomplete Network Datasets Using Simplified Graph Convolutional Neural Networks
title_sort exploiting weak ties in incomplete network datasets using simplified graph convolutional neural networks
publisher MDPI AG
series Machine Learning and Knowledge Extraction
issn 2504-4990
publishDate 2020-05-01
description This paper explores the value of <i>weak-ties</i> in classifying academic literature with the use of graph convolutional neural networks. Our experiments look at the results of treating <i>weak-ties</i> as if they were <i>strong-ties</i> to determine if that assumption improves performance. This is done by applying the methodological framework of the Simplified Graph Convolutional Neural Network (SGC) to two academic publication datasets: Cora and Citeseer. The performance of SGC is compared to the original Graph Convolutional Network (GCN) framework. We also examine how node removal affects prediction accuracy by selecting nodes according to different centrality measures. These experiments provide insight for which nodes are most important for the performance of SGC. When removal is based on a more localized selection of nodes, augmenting the network with both <i>strong-ties</i> and <i>weak-ties</i> provides a benefit, indicating that SGC successfully leverages local information of network nodes.
topic graph convolutional neural networks
weak ties
social networks
collective classification
url https://www.mdpi.com/2504-4990/2/2/8
work_keys_str_mv AT nedahbidoki exploitingweaktiesinincompletenetworkdatasetsusingsimplifiedgraphconvolutionalneuralnetworks
AT alexandervmantzaris exploitingweaktiesinincompletenetworkdatasetsusingsimplifiedgraphconvolutionalneuralnetworks
AT gitasukthankar exploitingweaktiesinincompletenetworkdatasetsusingsimplifiedgraphconvolutionalneuralnetworks
_version_ 1724656088896241664

Similar Items